Added Blake3 implementation using blake3 crate (#1613)

* Added Blake3 implementation using blake3 crate Signed-off-by: Prakhar Singhal <[email protected]> Added tests for blake3 implementation Signed-off-by: Prakhar Singhal <[email protected]> added/integrated blake3 function to crypto lib Signed-off-by: Prakhar Singhal <[email protected]> Added tests for blake3 in grammer tests Signed-off-by: Prakhar Singhal <[email protected]> Addded external dependency in cargo.toml and fixed formatting Signed-off-by: Prakhar Singhal <[email protected]> * minor bug fix Signed-off-by: Prakhar Singhal <[email protected]> --------- Signed-off-by: Prakhar Singhal <[email protected]>
kcl-lang · Sep 3, 2024 · bf8505d · bf8505d
1 parent fd9fe09
commit bf8505d
Show file tree

Hide file tree

Showing 13 changed files with 121 additions and 3 deletions.
diff --git a/compiler_base/3rdparty/rustc_span/Cargo.toml b/compiler_base/3rdparty/rustc_span/Cargo.toml
@@ -22,4 +22,5 @@ cfg-if = "0.1.2"
 tracing = "0.1"
 sha1 = { package = "sha-1", version = "0.10.0" }
 sha2 = "0.10.1"
-md5 = { package = "md-5", version = "0.10.0" }
+md5 = { package = "md-5", version = "0.10.0" }
+blake3 = "1.5.4"
diff --git a/compiler_base/3rdparty/rustc_span/src/lib.rs b/compiler_base/3rdparty/rustc_span/src/lib.rs
@@ -36,6 +36,7 @@ use std::rc::Rc;
 use std::str::FromStr;
 
 use md5::Digest;
+use blake3::Hash as Blake3;
 use md5::Md5;
 use sha1::Sha1;
 use sha2::Sha256;
@@ -602,6 +603,7 @@ pub enum SourceFileHashAlgorithm {
     Md5,
     Sha1,
     Sha256,
+    Blake3,
 }
 
 impl FromStr for SourceFileHashAlgorithm {
@@ -612,6 +614,7 @@ impl FromStr for SourceFileHashAlgorithm {
             "md5" => Ok(SourceFileHashAlgorithm::Md5),
             "sha1" => Ok(SourceFileHashAlgorithm::Sha1),
             "sha256" => Ok(SourceFileHashAlgorithm::Sha256),
+            "blake3" => Ok(SourceFileHashAlgorithm::Blake3),
             _ => Err(()),
         }
     }
@@ -643,6 +646,12 @@ impl SourceFileHash {
             SourceFileHashAlgorithm::Sha256 => {
                 value.copy_from_slice(&Sha256::digest(data));
             }
+            SourceFileHashAlgorithm::Blake3 => {
+                let mut hasher = Blake3::new();
+                hasher.update(data);
+                let result = hasher.finalize();
+                value.copy_from_slice(result.as_bytes());
+            }
         }
         hash
     }
@@ -663,6 +672,7 @@ impl SourceFileHash {
             SourceFileHashAlgorithm::Md5 => 16,
             SourceFileHashAlgorithm::Sha1 => 20,
             SourceFileHashAlgorithm::Sha256 => 32,
+            SourceFileHashAlgorithm::Blake3 => 32,
         }
     }
 }

diff --git a/kclvm/Cargo.lock b/kclvm/Cargo.lock
diff --git a/kclvm/runtime/Cargo.toml b/kclvm/runtime/Cargo.toml
@@ -21,6 +21,7 @@ regex = "1.5.5"
 md5 = "0.7.0"
 sha2 = "0.9.8"
 sha1 = "0.6.0"
+blake3 = "1.5.4"
 chrono = "0.4.19"
 ahash = "0.7.2"
 indexmap = "1.0"

diff --git a/kclvm/runtime/src/_kclvm.h b/kclvm/runtime/src/_kclvm.h
@@ -239,6 +239,8 @@ kclvm_value_ref_t* kclvm_crypto_sha384(kclvm_context_t* ctx, kclvm_value_ref_t*
 
 kclvm_value_ref_t* kclvm_crypto_sha512(kclvm_context_t* ctx, kclvm_value_ref_t* args, kclvm_value_ref_t* _kwargs);
 
+kclvm_value_ref_t* kclvm_crypto_blake3(kclvm_context_t* ctx, kclvm_value_ref_t* args, kclvm_value_ref_t* _kwargs);
+
 kclvm_value_ref_t* kclvm_crypto_uuid(kclvm_context_t* ctx, kclvm_value_ref_t* _args, kclvm_value_ref_t* _kwargs);
 
 kclvm_value_ref_t* kclvm_datetime_date(kclvm_context_t* ctx, kclvm_value_ref_t* _args, kclvm_value_ref_t* _kwargs);

diff --git a/kclvm/runtime/src/_kclvm.rs b/kclvm/runtime/src/_kclvm.rs
@@ -112,6 +112,7 @@ pub enum ApiFunc {
     kclvm_crypto_sha256,
     kclvm_crypto_sha384,
     kclvm_crypto_sha512,
+    kclvm_crypto_blake3,
     kclvm_crypto_uuid,
     kclvm_datetime_date,
     kclvm_datetime_now,

diff --git a/kclvm/runtime/src/_kclvm_addr.rs b/kclvm/runtime/src/_kclvm_addr.rs
@@ -113,6 +113,7 @@ pub fn _kclvm_get_fn_ptr_by_name(name: &str) -> u64 {
         "kclvm_crypto_sha256" => crate::kclvm_crypto_sha256 as *const () as u64,
         "kclvm_crypto_sha384" => crate::kclvm_crypto_sha384 as *const () as u64,
         "kclvm_crypto_sha512" => crate::kclvm_crypto_sha512 as *const () as u64,
+        "kclvm_crypto_blake3" => crate::kclvm_crypto_blake3 as *const () as u64,
         "kclvm_crypto_uuid" => crate::kclvm_crypto_uuid as *const () as u64,
         "kclvm_datetime_date" => crate::kclvm_datetime_date as *const () as u64,
         "kclvm_datetime_now" => crate::kclvm_datetime_now as *const () as u64,

diff --git a/kclvm/runtime/src/_kclvm_api_spec.rs b/kclvm/runtime/src/_kclvm_api_spec.rs
@@ -794,6 +794,10 @@
 // api-spec(c):    kclvm_value_ref_t* kclvm_crypto_sha512(kclvm_context_t* ctx, kclvm_value_ref_t* args, kclvm_value_ref_t* _kwargs);
 // api-spec(llvm): declare %kclvm_value_ref_t* @kclvm_crypto_sha512(%kclvm_context_t* %ctx, %kclvm_value_ref_t* %args, %kclvm_value_ref_t* %_kwargs);
 
+/// api-spec:       kclvm_crypto_blake3
+/// api-spec(c):    kclvm_value_ref_t* kclvm_crypto_blake3(kclvm_context_t* ctx, kclvm_value_ref_t* args, kclvm_value_ref_t* _kwargs);
+/// api-spec(llvm): declare %kclvm_value_ref_t* @kclvm_crypto_blake3(%kclvm_context_t* %ctx, %kclvm_value_ref_t* %args, %kclvm_value_ref_t* %_kwargs);
+
 // api-spec:       kclvm_crypto_uuid
 // api-spec(c):    kclvm_value_ref_t* kclvm_crypto_uuid(kclvm_context_t* ctx, kclvm_value_ref_t* _args, kclvm_value_ref_t* _kwargs);
 // api-spec(llvm): declare %kclvm_value_ref_t* @kclvm_crypto_uuid(%kclvm_context_t* %ctx, %kclvm_value_ref_t* %_args, %kclvm_value_ref_t* %_kwargs);

diff --git a/kclvm/runtime/src/crypto/mod.rs b/kclvm/runtime/src/crypto/mod.rs
@@ -1,5 +1,6 @@
 //! Copyright The KCL Authors. All rights reserved.
 
+extern crate blake3;
 extern crate md5;
 extern crate sha1;
 extern crate sha2;
@@ -167,6 +168,34 @@ pub extern "C" fn kclvm_crypto_sha512(
     panic!("sha512() missing 1 required positional argument: 'value'");
 }
 
+// blake3(value: str, encoding: str = "utf-8") -> str
+
+#[no_mangle]
+#[runtime_fn]
+pub extern "C" fn kclvm_crypto_blake3(
+    ctx: *mut kclvm_context_t,
+    args: *const kclvm_value_ref_t,
+    _kwargs: *const kclvm_value_ref_t,
+) -> *const kclvm_value_ref_t {
+    let args = ptr_as_ref(args);
+    let ctx = mut_ptr_as_ref(ctx);
+    if let Some(s) = args.arg_i_str(0, None) {
+        let mut hasher = blake3::Hasher::new();
+        hasher.update(s.as_bytes());
+        let result = hasher.finalize();
+
+        let mut hex = String::with_capacity(2 * Sha256::output_size());
+        use std::fmt::Write;
+
+        for byte in result.as_bytes() {
+            let _ = write!(&mut hex, "{byte:02x}");
+        }
+
+        return ValueRef::str(hex.as_ref()).into_raw(ctx);
+    }
+    panic!("blake3() missing 1 required positional argument: 'value'");
+}
+
 #[no_mangle]
 #[runtime_fn]
 pub extern "C" fn kclvm_crypto_uuid(

diff --git a/kclvm/sema/src/builtin/system_module.rs b/kclvm/sema/src/builtin/system_module.rs
@@ -1386,6 +1386,27 @@ register_crypto_member! {
         false,
         None,
     )
+    blake3 => Type::function(
+        None,
+        Type::str_ref(),
+        &[
+            Parameter {
+                name: "value".to_string(),
+                ty: Type::str_ref(),
+                has_default: false,
+                range: dummy_range(),
+            },
+            Parameter {
+                name: "encoding".to_string(),
+                ty: Type::str_ref(),
+                has_default: true,
+                range: dummy_range(),
+            },
+        ],
+        r#"Encrypt the string `value` using `BLAKE3` and the codec registered for encoding."#,
+        false,
+        None,
+    )
     uuid => Type::function(
         None,
         Type::str_ref(),

diff --git a/kclvm/tests/test_units/runtime/crypto/test_crypto.py b/kclvm/tests/test_units/runtime/crypto/test_crypto.py
@@ -38,6 +38,9 @@ def sha384(self, value: str) -> str:
     def sha512(self, value: str) -> str:
         return self.dylib.Invoke(f"crypto.sha512", value)
 
+    def blake3(self, value: str) -> str:
+        return self.dylib.Invoke(f"crypto.blake3", value)
+
     def test_md5(self):
         self.assertEqual(
             self.md5("The quick brown fox jumps over the lazy dog"),
@@ -77,6 +80,12 @@ def test_sha512(self):
             self.sha512(""),
             "cf83e1357eefb8bdf1542850d66d8007d620e4050b5715dc83f4a921d36ce9ce47d0d13c5d85f2b0ff8318d2877eec2f63b931bd47417a81a538327af927da3e",
         )
+
+    def test_blake3(self):
+        self.assertEqual(
+            self.blake3(""),
+            "af1349b9f5f9a1a6a0404dea36dcc9499bcb25c9adc112b7cc9a93cae41f3262",
+        )
 
 
 if __name__ == "__main__":

diff --git a/test/grammar/builtins/crypto/blake/blake3/main.k b/test/grammar/builtins/crypto/blake/blake3/main.k
@@ -0,0 +1,3 @@
+import crypto
+
+blake3 = crypto.blake3("ABCDEF")
diff --git a/test/grammar/builtins/crypto/blake/blake3/stdout.golden b/test/grammar/builtins/crypto/blake/blake3/stdout.golden
@@ -0,0 +1 @@
+blake3: 61c8c05f3e588c663cd9fbb1d7ad93604ed08cf335497095a020222cc9976cf1